Department of Physics and Astronomy
University of Heidelberg
Master thesis in Physics
submitted by
Manuel Gerken
born in Frankfurt am Main
2016
Gray Molasses Cooling of Lithium-6 Towards a
Degenerate Fermi Gas
This Master thesis has been carried out by Manuel Gerken
at the
Physikalisches Institut Heidelberg
under the supervision of
Prof. Dr. Matthias Weidemüller iv Kühlen mit grauer Melasse zur Realisierung eines entarteten Fermi Gases
Die vorliegende Arbeit beschreibt das Design, die Implementierung und die Charakteri- sierung einer Kühlung in grauer Melasse auf der D1 Linie von Lithium-6 Atomen. Durch diese zusätzliche Laser-Kühlung erreichen wir sub-Doppler Temperaturen für eine an- fänglich magneto-optisch gefangenes Gas und erhöhen seine Phasenraumdichte um einen Faktor von etwa 10. Kühlen mit grauer Melasse kombiniert die geschwindigkeitsabhängige Besetzung eines kohärenten Dunkelzustands mit einem sisyphusartigen Kühlprozess auf einer ortsabhän- gigen Energieverschiebung. Wir präsentieren Berechnungen und Analysen von bekleide- ten Energiezuständen von Lithium-6 in ein- und dreidimensionalen Polarisationsgradien- tenfeldern. Dieser erzeugen die ortsabhängige Energieverschiebung für die Kühlung mit grauer Melasse. Wir beschreiben die Planung und den Aufbau des Experiments, in dem 3.2×107 Atome innerhalb von 1 ms von 240 µK auf 42 µK herabgekühlt werden. Dies ent- spricht einer Einfangeffizienz von 80% gegenüber der anfangs in der magneto-optischen Falle hergestellten Probe. Wir messen die Auswirkungen von Frequenzverstimmung, Dau- er, Magnetfeld und Lichtintensitäten auf den Kühlprozess um optimale Parameter zu er- halten. Wir diskutieren, wie die erhöhte Phasenraumdichte die Übertragungseffizienz in die optische Dipolfalle verbessert. Die verbesserten Ausgangsbedingungen für evaporati- ves Kühlen ermöglichen es uns, ein stark entartetes Fermigas herzustellen, der Grundbau- stein für die Generierung einer suprafluiden Bose-Fermi-Mischung und die Untersuchung von Fermi Polaronen.
Gray Molasses Cooling of Lithium-6 Towards a Degenerate Fermi Gas
This thesis presents the design, implementation and characterization of gray molasses cooling on the D1 line of Lithium-6 atoms. With this approach we reach sub-Doppler temperatures for an initially magneto-optically trapped gas and increase its phase space density by a factor of approximately 10. Gray molasses cooling consists of a velocity-selective coherent population trapping of atoms at zero velocity in a dark state that works in combination with a Sisyphus-like cooling process in a spatially dependent light-shift. We present calculations and analysis of dressed state energies of Lithium-6 in one and three dimensional polarization gradient fields, which generate the spatially dependent energy shift of the gray molasses cooling. Design and setup for the experiment are described, in which 3.2 × 107 atoms are cooled from 240 µK to 42 µK in 1 ms. The atom number represents an 80% capture efficiency of the initial sample prepared in the magneto-optical trap. We measure and optimize the impact of frequency detuning, duration, magnetic field and beam intensities onto the cooling process. We discuss how the increased phase space density will improve the transfer efficiency into the optical dipole trap. The enhanced starting conditions for forced evaporative cooling will enable us to achieve a highly degenerate Fermi gas, the keystone to a double superfluid Bose-Fermi mixture and the investigation of Fermi polarons.
v
Contents
1 Introduction 1
2 Concept of gray molasses cooling 5 2.1 Principle of gray molasses cooling ...... 5 2.2 Λ configuration and coherent dark state ...... 7 2.3 Polarization gradient fields ...... 9 2.4 Gray molasses of Lithium-6 ...... 12 2.5 Calculations for Lithium-6 in gray molasses ...... 15 2.5.1 Description of light polarization ...... 15 2.5.2 Hamiltonian describing gray molasses ...... 17 2.5.3 Discussion of gray molasses in 1D Lin⊥Lin field ...... 20 2.5.4 Discussion of gray molasses in a 3D σ+-σ− field ...... 24
3 Implementation of gray molasses cooling of Lithium-6 27 3.1 Setup of the gray molasses ...... 27 3.1.1 Laser system and stabilization ...... 28 3.1.2 Optical setup for light preparation ...... 30 3.1.3 Optical setup at experimental chamber ...... 32 3.2 Characterization of gray molasses cooling ...... 33 3.2.1 Experimental sequence ...... 33 3.2.2 Measurements of gray molasses cooling ...... 35
4 Conclusion and further developments 45
Bibliography 53
Acknowledgment 63
vii
1 Introduction
Ultracold atomic gases yield an advantageous environment for the probing and simulation of few and many-body physics (Bloch et al., 2008). The features of controlling internal degrees of freedom, external motional states, the density of the gas, the potential shape and even the atomic interactions, of a cloud of atoms or a mixture of atoms, makes the field feasible for the investigation of a wide range of physical questions. One example is the simulation of magnetic spins in a condensed matter system where the interaction between the spins is of a long- range nature and spatially anisotropic. Such a system can be simulated with an ultracold gas of polar molecules where the strength of the electric dipole-dipole interaction can be controlled to uttermost accuracy (Pupillo et al., 2008; Quéméner and Julienne, 2012). Another example are Fermi impurities interacting with a degenerate Bose environment. This system simulates electrons moving in a lattice of positively charged atoms in a crystal. The interaction between the electron and the environment leads to a quasi-particle called "Fröhlich-polaron" with a different mass and energy compared to the bare electron (Fröhlich, 1954; Tempere et al., 2009). The last example is the generation of a double superfluid phase of a Bose- Fermi mixture. The historical approach to achieve this phase using Helium-4 and Helium-3 reaches a limit when decreasing the temperature. The density of Helium-4 drops to zero before Helium-3 reaches superfluidity (Edwards, 2013). Such a system can be simulated using ultracolt atomic gases. All of these three examples can be studied in an ultracold Bose-Fermi mixture of Cesium-133 and Lithium-6 under favorable conditions. A polar molecule of Cesium- 133 and Lithium-6 holds the strongest electric dipole moment in the singlet ground state of all alkali combinations leading to enhanced interaction effects (Aymar and Dulieu, 2005). The high mass ratio of Cesium-133 and Lithium-6 of 22.1:1 leads to simplifications in the theoretic description for Fermi polarons and double superflu- idity. Theories for static polarons, in the limit of infinitely heavy impurities, can directly be tested in the case of Cesium-133 impurities immersed in a Lithium-6
1 Chapter 1. Introduction
Fermi sea. The mass ratio is predicted to change the interaction energy between the two superfluids (Zhang et al., 2014). The study of all these fundamental properties of a strongly interaction Bose-Fermi gas will benefit from a favorable scattering resonances between Lithium-6 and Cesium-133, leading to controllable intra and inter species interactions (Tung et al., 2013; Repp et al., 2013; Pires et al., 2014; Ulmanis et al., 2015). A long journey of progress in technical and physical understandings of 40 years lead to the point where we can address these questions in ultracold atomic systems. The proposal by Hänsch and Schawlow (1975) to cool a vapor of atoms by intense quasi-monochromatic laser light in 1975 was the beginning of laser cooling, a mile- stone in ultracold atomic physics. In the following years laser cooling and trapping methods were developed and awarded with the Nobel Prize in 1997 (Phillips, 1998; Chu, 1998; Cohen-Tannoudji, 1998). The realization of Bose-Einstein condensation (BEC) in diluted atomic gases in 1995 (Davis et al., 1995; Bradley et al., 1995; An- derson et al., 1995) was the first experimental generation of a degenerate quantum gas and was honored with the Nobel prize in 2001 (Cornell and Wieman, 2002; Ketterle, 2002). Due to the ultralow temperature, a dilute Bose gas with a phase space density of approximately one can be produced. In such a gas a macroscopic part of the ensemble occupies the lowest energy state of the system. This leads to a macroscopic quantum wavefunction. A few years after the generation of BEC the first degenerate Fermi gas was realized by laser cooled atoms (DeMarco and Jin, 1999; Schreck et al., 2001; Truscott et al., 2001; Hadzibabic et al., 2002). The discovery of magnetic Feshbach resonances in 1998 marked another milestone in the field of ultracold atomic physics (Inouye et al., 1998; Courteille et al., 1998). Such scattering resonances allow to control and tune the complete scattering process at ultracold temperatures, where it is described by the s-wave scattering length a. This ability of tuning the interaction between two different atoms or two spin components of the same atomic species opened the door for the observation of many new effects. The first realization of a BEC out of weakly bound molecules (a > 0), formed from two fermionic components, was achieved in 2003 (Jochim et al., 2003; Zwierlein et al., 2003; Greiner et al., 2003). For weakly attractive interactions (a < 0) pairing mechanism of a two component Fermi gas could be investigated (Chin et al., 2004; Schunck et al., 2008). This mechanism describes low temperature superconductivity and was predicted by the Bardeen-Cooper-Schrieffer theory (BCS) (Bardeen et al., 1957).
2 Ultracold gases with tunable interactions have been applied to address polaron physics and double superfluidity. Attractive and repulsive Fermi polarons have been detected for impurities in a highly degenerate, single-species fermionic sam- ple by changing the internal state of the impurity from a non-interacting into an interacting state while tuning the interactions between bath and impurity using Feshbach resonances (Schirotzek et al., 2009; Kohstall et al., 2012; Koschorreck et al., 2012). In 2016 ultrafast dynamics of impurities immersed in a single species Fermi sea were investigated (Cetina et al., 2016). The amplitude damping of the coherent dynamics was directly connected to the degeneracy of the bath. A high degeneracy is thus necessary for precise observation of the collective response. Only in 2016 the first observations of attractive and repulsive Bose polarons in a BEC were reported (Jørgensen et al., 2016; Hu et al., 2016). In Bose-Fermi mixtures the first double superfluid was realized by Ferrier-Barbut et al. (2014) in a mixture of Lithium-7 and Lithium-6 and by Roy et al. (2016) in a mixture of Ytterbium-174 and Lithium-6. Coherent energy exchange between both species was detected by driving center-of-mass oscillations. Yao et al. (2016) found visual evidence for dou- ble superfluidity in a mixture of Potassium-41 and Lithium-6 by rotating the gas and therefore creating simultaneously existing vortex lattices. Neither of these effects have been realized in a mixture of Cesium-133 and Lithium-6. While a BEC of Cesium-133 (Weber et al., 2003; Kraemer et al., 2004; Pires et al., 2014) and superfluidity of Lithium-6 are nowadays routinely realized in many groups (Zwierlein et al., 2006b; Sidorenkov et al., 2013; Ferrier-Barbut et al., 2014; Yao et al., 2016; Roy et al., 2016), it has so far not been possible to reach superfluidity of either lithium or cesium in the presence of the other species. In our group we are able to produce degenerate Lithium-6 and Cesium-133 sam- ples separately. The Cesium-133 BEC sample contains 5×104 atoms with a conden- sate fraction of approximately 70% at a magnetic field of 22.8 G (Pires, 2014). For Lithium-6 we achieve 5×104 atoms with a condensate fraction of 70% at a magnetic field of 690 G (Repp, 2013). However superfluid Lithium-6 on the BCS side (a < 0) has not been realized in our experiment. The sample temperature to achieve super- fluidity for a two component Fermi gas at negative scattering length for the specif harmonic trapping potential in our experiment is given by approximately
T . 10−10N 1/3K. (1.1)
Here T is the sample temperature and N is the number of atoms. In order to reach
3 Chapter 1. Introduction superfluidity, either the temperature has to be reduced or the number of atoms has to be increased. The latter can be achieved by enhancing phase space density of the Lithium-6 cloud before transferring it into the optical dipole trap to improve starting conditions for forced evaporative cooling (Ketterle and van Druten, 1996).
Due to the unresolved hyperfine splitting of the D2 line of Lithium-6 the sub- Doppler cooling process, working in the magneto-optical trap (MOT) of, for exam- ple sodium (Lett et al., 1988) and other species, does not work here. This leads to a MOT temperature that is limited by the Doppler temperature which is around 140 µK for Lithium-6. Two approaches have been used to reach lower tempera- tures. Narrow-line cooling (Duarte et al., 2011; McKay et al., 2011) applied on 2 2 the 2 S1/2 → 3 P3/2 transition of Lithium-6 at 323 nm, provides temperatures of ≈ 60 µK, increasing the phase space density of the MOT by one order of magnitude. Due to the narrow linewidth of the uv transition of Γ = 2π × 754 kHz the Doppler temperature decreases to 18 µK. Another method is the one we present here. Gray molasses cooling on the D1 line is a process closely related to Sisyphus cooling. It is based on the concept of velocity selective coherent population trapping (VSCPT) (G. Grynberg, 1994; Weidemüller et al., 1994). Using this method temperatures down to 60 µK, 40 µK, 20 µK and 6 µK have been reached for Lithium-7, Lithium-6, Potassium-40 and Potassium-39, respectively (Grier et al., 2013; Burchianti et al., 2014; Rio Fernandes et al., 2012; Salomon et al., 2013). In this way a temperature decrease of approximately one order of magnitude can be realized for Lithium-6 while maintaining the MOT atom number and sample density. This thesis covers calculations, design, implementation, characterization and fu- ture plans of gray molasses cooling in the existing experimental setup. In chapter 2 we give an overview over the working principles of gray molasses cooling by first presenting a simplified cooling scheme in a three-level Λ system and later extending the model to the case of Lithium-6. We present calculations for Lithium-6 in a one and three dimensional gray molasses field with two different polarization configu- rations. Chapter 3 presents the experimental setup, implementation as well as the characterization of the gray molasses cooling. In the final chapter we summarize the findings and give an outlook on experiment prospects coming into reach by gray molasses cooling.
4 2 Concept of gray molasses cooling
In this chapter we discuss the basic principles of gray molasses cooling on the D1 line of Lithium-6. At first we give a brief overview over the cooling process in a gray molasses in section 2.1. In section 2.2 we take a closer look at a 3-level Λ configuration and map out three basic points to explain the cooling process. We examine the effect of a light polarization gradient field onto the system and discuss two possible field configurations. Section 2.4 introduces Lithium-6 and discusses the effect of the gray molasses. In section 2.5 we present calculations for Lithium-6 in two different polarization gradient field configurations in one and three dimensions, which gives us a deeper understanding of the cooling mechanism and its dependence on the Raman detuning, the phase difference between two frequency components and the magnetic field.
2.1 Principle of gray molasses cooling
The main idea of gray molasses cooling can be described as follows: An atom propagates in a blue detuned light field with spatially dependent polarization. The field couples atomic states, such that the ground state manifold is divided into bright states |ψB⟩ and dark states |ψD⟩. The dark states experience no energy shift due to the light field. The bright states however experience a spatially varying light shift induced by the polarization gradient of the field. Figure 2.1 shows a schematic picture of bright state |ψB⟩ energy and dark state |ψD⟩ energy along the z axis in the light field. Now we consider an atom with velocity v propagating along the light field in positive z direction. In the bright state, the atom will exchange kinetic and potential energy when climbing a potential hill. At the point of highest light shift it will have the lowest velocity. The probability of the atom to absorb a photon is proportional to the light shift, thus it is more probable for an atom to get excited at the top of the potential hill than in the potential valley. After absorbing a photon, the atom will either decay back into the bright state where it will again absorb a
5 Chapter 2. ConceptD. Rio of gray Fernandes molasseset al. cooling
E 훿 F'=7/2 2 P1/2 훿 2 155.3 MHz
F'=9/2
Cooling ψB 770.1 nm 0 Repumping
ψD
z F=7/2 λ/4 λ/2 3λ/4
0 2 Fig. 1: Gray Molasses scheme. On a F → F = F or S1/2 1285.8 MHz Figure 2.1: A positiveF → F detuned0 = F − 1 optical optical transition transition with in a spatialpositive dependent detuning, polarization fieldthe splits ground the ground state splits states into into a dark a bright and a bright state | manifoldψB⟩ and a dark state |ψD⟩ withmanifold. positive Kinetic energy, energy shown as is transfered|ψDi and |ψ toBi potentialrespectively. energy when an F=9/2 In the presence of a polarization gradient, the bright state atom climbs the potential hill of the bright state. When pumpedFig. 2: from Level scheme for the D transition of 40K and energy is spatially modulated. Like in Sisyphus cooling, 1 the bright to the dark state the atom looses energy. Atoms transfertransitions from used for gray molasses cooling. The laser de- energy is lost when an atom in |ψ i climbs a potential the dark to the bright state due to velocityB selective coupling.tuning Taken from the cooling/repumping transitions is δ and hill before being pumped back into the dark state |ψ i. D the detuning from the off-resonant excited hyperfine state fromMotional (Rio Fernandes coupling et between al., 2012|ψ).i and |ψ i occurs prefer- D B F 0 = 9/2 is δ (see text). entially at the potential minima. 2 photon, or it will decay into the dark state. The decay into the dark state causes the atom to loseon energy the F on→ theF 0 order= F ( ofF the→ F bright0 = F state− 1) energy optical shift. tran- As mentioned above, two mechanisms can lead to the sition. For any polarization of the local electromagnetic departure from the dark state. The first one is the mo- An atom populatingfield, the the ground dark state state manifold can transfer possesses to the one bright (two) dark statetional via velocity coupling Vmot due to the spatial variations of the selective coupling.states The which probability are not optically of the process coupled is to proportional the excited state to thedark velocity state internalof wave-function induced by polarization by the incident light [12,16]. When the laser is detuned to and intensity gradients. The second one is the dipolar the atom and inversely proportional to the light shift of the bright state (Dalibard the blue side of the resonance, the ground state manifold coupling Voff via off-resonant excited hyperfine states. A and Cohen-Tannoudjisplits, into1989 dark). This states process which are leads not to affected velocity by selective light and coherentrough estimate pop- shows that Vmot ' ~kv, where v is the bright states which are light-shifted to positive energy by velocity of the atom and k the wave-vector of the cool- ulation trapping (VSCPT) (Arimondo, 1992). Any atom with a velocity other than −1 an amount which depends on the actual polarization and ing light, while Voff ' ~Γ (Γ/δ2) I/Isat, where Γ is the zero will take partintensity in the of cooling the laser cycle field until (see fig. reaching 1). zero velocity. Atomslifetime at of rest the excited state, I the light intensity, Isat the will accumulate inWhen the dark the state. atom is The in a process bright state, is therefore it climbs not up thelimited hill bysaturation the single intensity and δ2 the detuning to off-resonant ex- of the optical potential before being pumped back to the cited state. Comparing the two couplings, we see that the photon recoil energydark state which near is the the top limit of of the Sisyphus hill. The cooling kinetic energy (Castin of etmotional al., 1991 coupling). is significant in the high velocity regime Thus temperaturesthe below atom is the thus single reduced photon by an recoil amount temperature of the order of the v & Γ/k (Γ/δ2) I/Isat. In our case, the off-resonant level 0 height of the optical potential barrier. The cooling cycle F = 9/2 (see fig. 2) is detuned by δ2 = 155.3 MHz+δ from 2 2 0 ~2 2 the cooling transition | S1/2,F = 9/2i → | P1/2,F = is completed near the potentialk minimap by a combination T = , 7/2i. For(2.1)I ' I , motional coupling dominates for of motional couplingrecoil and optical2mk excitation to off-resonant sat hyperfine states. B T & 50 µK, meaning that both processes are expected to 40 be present in our experiments. In general, the transition where k is the photonWe implement wave number 3D gray and molassesm is cooling the atom in mass,K on the can be achieved p rate between |ψ i and |ψ i induced by motional coupling D1 transition (see fig. 2). In alkali atoms, the P1/2 excited D B (Esslinger et al.,level1996 manifold). However, has only off-resonant two hyperfine coupling states, and which imperfections are Vmot and in the the off-resonant coupling Voff are both maximal when the distance between the dark and bright manifolds better resolved than their P3/2 counterparts. These facts allow for less off-resonant excitation and a good control of is smallest, which favors transitions near the bottom of the cooling mechanism. A first laser beam (cooling beam) the wells of the optical lattice. 6 2 2 0 40 is tuned to the | S1/2,F = 9/2i → | P1/2,F = 7/2i In K, the simplified discussion presented so far must transition with a detuning δ > 0. A second laser beam be generalized to the case involving many hyperfine states 2 (repumping beam) is tuned to the | S1/2,F = 7/2i → (10 + 8). However, the essential picture remains valid. In- 2 0 | P1/2,F = 7/2i transition with the same detuning δ. deed, by numerically solving the optical Bloch equations
p-2 2.2. Λ configuration and coherent dark state polarization field can disturb the process and limit the achievable temperature. In some systems temperatures above the recoil limit but still an order of magnitude beneath the Doppler limit can be reached (Grier et al., 2013).
2.2 Λ configuration and coherent dark state
To describe the emergence of dark and bright states and velocity selective coupling in a simple picture we here introduce the Λ three-level system. The Λ three-level model is shown in figure 2.2. It describes a system with two ground states |1⟩ and |2⟩ and one excited state |3⟩, with energies ϵ1, ϵ2 and ϵ3, respectively. States | ⟩ | ⟩ 1 and√3 are coupled via an oscillator with frequency ω1 and Rabi frequency Ω1 = Γ I1/2Isat, where the frequency is detuned by δ1 from resonance, Γ is the linewidth of state |3⟩ and I1 the intensity of the light field. Isat is the saturation intensity (Weidemüller and Zimmermann, 2003) of the atomic transition with
1 Γ~ω3 I = 1 , (2.2) sat 12 π2c2 | ⟩ | ⟩ where c is the speed of light. States 2 and 3 are coupled by an oscillator√ with frequency ω2, detuned by δ2 from resonance and Rabi frequency Ω2 = Γ I2/2Isat.
The relative detuning of ω1 and ω2 is ∆ = δ1 − δ2. Diagonalizing the Hamiltonian of the system will give us the dressed states, describing the dark and the bright state. Under several conditions the system simplifies dramatically. First we assume no difference in energy of state |1⟩ and |2⟩ meaning ϵ1 = ϵ2. Furthermore we assume the so called Raman condition ∆ = 0.
We thus have ω1 = ω2. The Hamiltonian of this complete system H can be divided into the Hamiltonian without light interaction H0 and the coupling Hamiltonian V . In the rotating frame the Hamiltonian H can be written as: 0 0 Ω1 ~ H = H + V = 0 0 Ω , (2.3) 0 2 2 Ω1 Ω2 −2δ1 where H0 = −~δ1 |3⟩⟨3| is the Hamiltonian of the system without light coupling and −~δ1 is the energy of the excited state. The coupling Hamiltonian is given by
~Ω ~Ω V = 1 |1⟩⟨3| + 2 |2⟩⟨3| + c.c. (2.4) 2 2
7 Chapter 2. Concept of gray molasses cooling
Δ
Figure 2.2: Three-level scheme coupled by two light fields with Rabi frequencies Ω1 and Ω2 detuned from resonance by δ1 and δ2, respectively. Adapted from (Grier et al., 2013).
It describes the coupling of states |1⟩, |2⟩ and |3⟩ by the light fields ω1 and ω2 where
ω1 = ω2 but Ω1 ≠ Ω2. After diagonalizing the Hamiltonian, we get two new eigenstates in the dressed state picture √ 1 |ψD⟩ = (Ω1 |1⟩ − Ω2 |2⟩), (2.5) 2 2 Ω1 + Ω2
√ 1 |ψB⟩ = (Ω1 |1⟩ + Ω2 |2⟩). (2.6) 2 2 Ω1 + Ω2
In order to determine whether the states couple to the light field or experience a light shift, we examine the effect of the coupling Hamiltonian on them. For state
|ψD⟩ we get V |ψD⟩ = 0. This means that state |ψD⟩ does not couple to the light | ⟩ field and does not experience a light√ shift. Consequently we call ψD the dark state. | ⟩ ⟨ | | ⟩ ~ 2 2 | ⟩ For ψB , we get 3 V ψB = 2 Ω1 + Ω2. This shows that state ψB couples to (Ω2+Ω2) the excited state |3⟩ via the light field. It experiences a light shift 1 2 and δ1 we therefore call it the bright state. For a positive (blue) detuning δ1 > 0 the light shift is positive, while for a negative (red) detuning δ1 < 0 the light shift is negative. Since |ψB⟩ and |ψD⟩ are orthogonal eigenstates in the basis of H, they do not couple, so that ⟨ψB| H |ψD⟩ = 0.
We now introduce velocity dependent coupling between |ψD⟩ and |ψB⟩. This was described in (Papoff et al., 1992) for the Λ three-level scheme. We will here briefly summarize the most relevant results. For this we include the kinetic energy term in H pˆ2 the full Hamiltonian = 2m + H, where H is the Hamiltonian from equation 2.3 and pˆ is the momentum operator.The coupling matrix element between the dark
8 2.3. Polarization gradient fields and the bright state yield
⟨ | H | ⟩ −~ 2Ω1Ω2 kp ψB ψD = 2 2 , (2.7) Ω1 + Ω2 m where p is the momentum of the atom and k is the wave number of the transitions. This shows a coupling depending on the velocity of the atom v = p/m. Papoff et al. (1992) calculate in first order perturbation theory the coupling probability ( ) 2 Ω1Ω2 p δ1 P| ⟩→| ⟩ = 2 k δ , (2.8) ΨD ΨB 2 2 ~ 1 2 2 Ω1 + Ω2 (Ω1 + Ω2) which describes the velocity selective coupling of an atom with momentum p from the dark state |ψD⟩ to the bright state |ψB⟩. Therefore the transition probability
δ1 is proportional to the velocity square. The last factor 2 2 describes the inverse (Ω1+Ω2) of the light shift of the bright state. δ1 describes the detuning from resonance, thus the coupling will be strongest at small light shifts and high velocity. This shows that all the slow atoms accumulate in the dark state |ψD⟩. The fast atoms are involved in the cooling cycle until the velocity is zero such that the motional coupling vanishes. With this model we can describe the following: An atom with high kinetic energy in the dark state |ΨD⟩ will couple to the bright |ψB⟩ state at a point close to the potential minimum of the bright state. Then moving in the light field it will climb the potential hill. After an excitation by photon absorption, it will decay back into the ground state manifold. If the atom decays into the bright state the process is repeated. If it decays into the dark state the atom looses kinetic energy on the order of the bright state light shift and is thus cooled.
2.3 Polarization gradient fields
To understand the origin of a spatial dependent bright state energy shift we will here discus the influence of a polarization gradient light field on an atom. For this we consider a system with a ground state of total angular momentum F = 1/2 which we denote |F = 1/2⟩ and an excited state of total angular momentum F ′ = 3/2 which we denote |F ′ = 3/2⟩. The states then consist of degenerate sub-
{− 1 1 } ′ states described by the magnetic quantum numbers mF = 2 , 2 and mF = {− 3 − 1 1 3 } 2 , 2 , 2 , 2 which are the projections of the total angular momentum. When considering optical dipole transitions, not all transitions are allowed. The selection
9 Chapter 2. Concept of gray molasses cooling